Menu audit

What the audit covers

1. A verified plate cost for every dish, built from your recipe specs and your last three months of invoices, with yield and trim loss loaded in.
2. Sales mix analysis across a minimum of eight weeks of EPOS data, cleaned for open keys and modifier errors.
3. The full matrix, with every dish placed as a star, plowhorse, puzzle or dog, by section rather than across the whole menu.
4. Your weighted gross profit, which is usually several points away from the figure people expect.
5. A ranked action list, ordered by annual cash impact and flagged for how difficult each change is to make.
6. A ninety minute session going through it with you and your head chef, on a call or in person.

To see what comes back, there is a sample menu profitability analysis, worked on a fictional restaurant: every dish costed, placed on the matrix and ranked by where to act first.

The number you are working from is probably wrong

Most kitchens have a food GP figure they quote. It comes off the P&L, it is a single number for the whole operation, and it is the number people manage against. The trouble is that it tells you almost nothing about which dishes are responsible for it.

There are two gaps worth separating. The first is between theoretical and actual food cost. Theoretical is what the menu should return if every plate went out exactly to spec. Actual comes off your stock take. Within a point of each other and your control is genuinely good. Two to three points apart is normal. Four or more and something structural is happening, and it is almost never the thing people assume.

The second gap is between your headline GP and your weighted GP. Weighted GP is the number your accounts actually see, because it reflects what people ordered rather than what was theoretically available to order. A menu where the strongest margins sit on dishes nobody chooses will post a healthy average and a disappointing bank balance. Sorting out which of those two gaps you have, and how much of it there is, is the first thing an audit settles.

What usually turns up

Every menu is different, but the same handful of findings come back often enough to be worth naming.

• Specifications that stopped matching the kitchen. A spec is written at launch, the dish evolves across eighteen months of chefs improving it, and nobody updates the document. The plate cost in your model is costing a dish you no longer serve.
• Protein costed at invoice price instead of yielded price. A whole fish at twelve pounds a kilo that breaks down to forty five percent usable fillet costs you nearly twenty seven pounds a kilo on the plate. Costings built on the invoice understate every protein on the menu.
• A best seller quietly subsidised by everything else. It sells because it is good and it is priced where people expect it, and it returns less cash than the dishes beside it. It is rarely the right call to delist it. It is usually the right call to rebuild what is on the plate.
• EPOS data that hides a third of the menu. Open keys, misused modifiers, specials rung through a generic button. If the till cannot tell you what sold, no analysis built on it is worth having, and cleaning that structure is often the first thing that pays for itself.
• High margin dishes in the places nobody reads. Menu position is not decoration. A dish that earns well and sits at the bottom of a long section is a positioning problem with a very cheap fix.

Why the matrix runs by section, not across the whole menu

Scoring an entire menu on one scale produces a predictable and useless answer: starters look unprofitable and main courses look like stars, because a nine pound plate cannot return the cash a twenty eight pound plate does. All you have measured is price.

Running the matrix by section compares dishes against the ones they genuinely compete with, which is the only comparison a guest ever makes. Nobody chooses between a starter and a dessert. They choose between three starters. Within a section, a dish that returns below the section average on cash margin while taking a healthy share of sales is a real finding and an actionable one. Across a whole menu, the same dish disappears into the noise.

You can see the mechanics of this on the food cost calculator, which runs the same two axes on a single dish. The audit does it across every line and weights the result by what actually sold.

What I need from you

• Current menus, all of them, including bar snacks and set lunch
• EPOS sales mix export, eight weeks minimum, twelve preferred
• Recipe specifications, in whatever state they are in
• Three months of supplier invoices or a price list export
• About an hour of your head chef's time, to talk through how the dishes are actually built

None of that needs tidying up first. Specifications written on the back of a prep list are more useful than specifications that do not exist, and the state the data arrives in is itself a finding. If your EPOS export is unreadable, that is worth knowing before you spend money on a menu redesign built on it.

The report, and why it is ordered the way it is

What comes back is a ranked list, not a document to admire. Every recommendation carries two things: what it is worth in cash across a year at your current volumes, and how hard it is to do. Those two together decide the order.

That matters because the biggest number on the list is often the hardest change, and starting there is how this work stalls. Re-specifying a dish your head chef has cooked for six years is a conversation. Moving that dish four lines up the section is an afternoon. Both are on the list; the order tells you which to do in week one and which needs planning. Nothing on it says "increase prices" as a general instruction, because that is not a finding, it is an absence of one. Where a price does need to move, the pricing guide sets out how the number is arrived at.

What the fee depends on

Quoted on scope rather than from a price list, because an audit of a fourteen dish restaurant and an audit of a pub running six live menus are not the same job. What moves the number is the number of dishes, the number of sites, and whether written specifications already exist or have to be built from scratch. That last one moves it most: building specs from scratch is the bulk of the work on a site that has never had them.

You get the figure before any work starts, and the fee is credited in full against a menu redesign commissioned afterwards. Timing is set against your own date when the quote goes out. The work runs remotely, from your figures and your specifications.

When it is not worth doing

An audit is diagnosis. If you already know exactly which dishes are the problem and you simply want them rebuilt, you want a redesign, and paying to be told what you already know is a waste. If your costings are current, your specs are written and your menu is short, there may be very little in it, and I would rather say so at the quote stage than at the report stage.

It is also not a food review. I am not there to judge the cooking, and in most cases it is good. The audit is about what the menu earns, not what it tastes like. And it does not come with a projection of what you will gain, because the size of the prize is not knowable until the gap has been measured. Measuring it is the job.

The audit scores every dish, and the findings read best section by section: menu profitability by section. What the work looks like end to end: a sample analysis. How the work is engaged: the terms of service.

Menu audit FAQs